In autonomous racing, vehicle control is critical to ensure the vehicle operates at the handling limits during aggressive maneuvers. This work presents a hierarchical control framework for autonomous racing, which consists of feedback linearized model predictive control (FBL-MPC) at the high level, together with model-based yaw rate control and powertrain model-based acceleration control at the low level. A reduced-order model is used for computationally efficient prediction, and the feedback linearization enables a linear MPC that is suitable for real-time optimization. Steady-state lateral slip effects are explicitly introduced into high-level prediction, and low-level yaw rate control helps mitigate the effects of model mismatch and unmodeled dynamics. The proposed controller has been implemented and experimentally validated on the full-scale IAC AV-24 autonomous race car during the Indy Autonomous Challenge at WeatherTech Raceway Laguna Seca, achieving stable operation and a lap time of 90.5 s. Additional track testing compares the effects of different prediction models and cost functions. The results demonstrate that the proposed controller framework achieves competitive performance in the racing scenario.

The paper addresses the critical need for a robust testing methodology to assess control performances of off-road vehicles with electrified powertrains due to the challenges posed by stringent emission regulations. The main contribution of this paper lies in the development of an innovative real-time hardware-in-the-loop (HIL) simulation platform tailored for a next-generation battery hybrid electric wheel loader proposed for John Deere. This platform integrates a unique vehicle power management (VPM) strategy, component-level controllers, and physics-based powertrain component models to simulate vehicle operations. Through this new HIL simulation platform, the paper demonstrates the potential for achieving an over 10% reduction in fuel consumption with the proposed next-generation battery hybrid electric wheel loader. The effectiveness of the platform is further validated through in-field testing, aligning the fuel efficiency capabilities and VPM strategy performance with the simulated results in the HIL environment.

This paper proposes a multiple-input multiple-output (MIMO) nonlinear feedback linearization controller to enhance load transient performance in a 4.5L off-road diesel engine equipped with an e-booster (electrically driven compressor). The engine's air-path subsystem is strongly coupled and highly nonlinear due to thermodynamic interactions between pressures and flows, making traditional single-input single-output (SISO) strategies less effective, especially during transients involving additional degrees of freedom such as e-booster actuation. A six-state mean value engine model based on physics is developed and used to design a non-linear MIMO controller that tracks key outputs: air-fuel ratio (AFR), engine speed and diluent-air ratio (DAR, representative of % EGR). Feedback linearization is applied to handle nonlinearities and decouple input-output dynamics, with a feedback PI controller added to improve robustness. The approach is compatible with production-level sensors, enabling practical implementation. The controller was evaluated during a simulated load transient from 100 Nm to 500 Nm in 0.3 seconds at constant speeds of 1200, 1600, and 1800 rpm, reflecting real-world off-road operating conditions. Compared to other linear control strategies, the proposed controller consistently achieved lower engine speed droop, improved DAR tracking, and minimized AFR dip.

This paper presents a novel approach for predicting the simulation-to-real-world (sim2real) transferability of synthetic-trained vision models by analyzing foreground image quality. Image quality assessment (IQA) techniques are applied to paired synthetic and real-world image subsets to quantify visual similarity and identify feature-level limitations that correlate with transfer performance. Among the evaluated metrics, the Structural Similarity (SSIM) index and Complex Wavelet SSIM (CW-SSIM) index showed consistent trends, where higher IQA scores corresponded to improved sim2real detection accuracy–most notably with increases of approximately 0.10 (SSIM) and 0.15 (CW-SSIM). These metrics effectively captured contrast, structure, and luminance similarities, offering a practical proxy for assessing digital twin fidelity. To support reproducibility and broader use, we introduce the Synthetic Image Quality Analysis Calculator (SIQAC), an open-source tool for automated IQA evaluation and sim2real potential prediction across classifiers and object detectors. Additional experiments demonstrated that the IQA-based approach generalizes to real2sim scenarios using zero-shot object detectors. This work bridges concepts from the human visual system and compression analysis to provide a lightweight, interpretable method for early-stage validation of virtual autonomy pipelines.

Accurate estimation and prediction of engine gas exchange system and in-cylinder conditions are critical for spark-ignited engine control and diagnostic algorithm development. In this paper, a physically-based, control-oriented model for a 2.8 l turbocharged, variable valve timing (VVT) and low pressure (LP) exhaust gas recirculation (EGR)-utilizing SI engine was developed. The model includes the impact of modulation to any combination of 10 actuators, including the throttle valve, compressor bypass valve, fueling rate, waste-gate, LP EGR valve, number of deactivated cylinders, intake valve open (IVO) timing, intake valve close (IVC) timing, exhaust valve open (EVO) timing and exhaust valve close (EVC) timing. The accuracy of the model in capturing engine dynamics was demonstrated by validating it against high-fidelity engine GT-Power simulation results for various drive cycles, particularly emphasizing elevated loads. In comparison to the open literature, novel contributions of the effort described in this paper includes in-cylinder gas composition modeling and turbine-out pressure estimation.

This paper proposes an innovative next-generation wheel loader derived from a production series electric wheel loader by John Deere. The proposed wheel loader features a series hybrid electric powertrain with an energy storage system and an electrically-boosted turbocharged diesel engine. A detailed system design inclusive of powertrain configuration and control development including a novel heuristics-based vehicle power management is presented. A simulation model was created to evaluate the potential fuel savings of the proposed wheel loader, revealing a fuel saving potential of over 10% when compared to the baseline configuration. Subsequent in-field testing of an actual demo wheel loader verified its ability to achieve over 10% fuel savings, thereby confirming the simulation outcomes, demonstrating the promise of the proposed hybrid powertrain, and validating the efficacy of the control system developed in this research.

Off-road heavy-duty diesel engines are equipped with complex aftertreatment systems to meet the stringent EPA Tier 4 emission standards. Traditionally, the thermal management of the aftertreatment system, aimed at efficiently reducing tailpipe emissions, involves controlling engine exhaust flow and temperature. However, this approach often leads to inefficient engine operation, especially in the low to mid-load regions, resulting in higher fuel consumption. Prior studies have highlighted the potential of cylinder deactivation (CDA) in reducing fuel consumption and increasing the exhaust temperature for thermal management in on-road applications. As the significance of controlling greenhouse gas (GHG) emissions from off-road machinery comes into focus, this study aims to experimentally demonstrate the fuel efficiency benefits and efficient aftertreatment thermal management achieved through CDA during high-speed operation on a Tier 4 level off-road heavy-duty diesel engine. In a typical off-road duty cycle, such as Non-Road Transient Cycle (NRTC), about 82.5% of cycle's energy is produced in the engine speed range of 1750-2200 rpm, prompting investigation into the impact of CDA during high-speed operations. CDA results in airflow reductions and increased exhaust gas temperatures due to lower air-fuel ratio (AFR). The reduced airflow operation minimizes pumping work, allowing for higher open cycle efficiency, and thereby translating into lower fuel consumption. The study reveals a new finding: fuel benefits from CDA extend over a larger load range (up to 8.3 bar BMEP) during high-speed operation in off-road heavy-duty engines compared to findings in on-road heavy-duty engines. This phenomenon can be attributed to sufficient oxygen inducted during CDA operation, resulting from similar turbocharger speeds compared to all-cylinder firing operation, especially at high-loads, thereby maintaining particulate matter (PM) concentration within prescribed constraints. Steady-state test results demonstrate a reduction in fuel consumption from 33.7% at 0.4 bar to 1.4% at 8.3 bar BMEP, at 2100 rpm engine speed.

This paper describes and demonstrates a model-based sensor selection and controller design framework for robust control of air–fuel-ratio, air flow and EGR flow for turbocharged stoichiometric engines using low pressure EGR, waste-gate turbo-charging, intake throttling and variable valve timing. Model uncertainties, disturbances, transport delays, and sensor- and actuator-characteristics are considered in this framework. Based on the required control performance and candidate sensor sets, the framework synthesizes an H∞ feedback controller and evaluates the viability of the candidate sensor set through analysis of the singular structured value μ of the closed-loop system in the frequency domain. The framework can also be used to understand if relaxing the controller performance requirements enable the use of a simpler (less costly) sensor set. The sensor selection and controller co-design approach is applied here, for the first time, to turbo-charged engines using exhaust gas circulation. High fidelity GT-Power simulations are used to validate the approach. The computed optimal controllers were able to maintain the AFR within 14.7 ± 0.3, control the torque within 8.2% error and the EGR ratio within 1.2% error at different drive cycles with sensor inaccuracies and actuator dynamics.

This paper demonstrates a multiple-input multiple-output (MIMO) controller design framework and a controller switching algorithm for MIMO controllers in their state-space form, which together achieve robust, efficient control of turbocharged lean-burn engines over a wide operating space. The controller design framework requires a linearized plant model, and uses the [Formula: see text]-synthesis and DK-iteration algorithms while considering state and output uncertainties and actuator bandwidths to synthesize a robust [Formula: see text] controller. A controller switching methodology using slow-fast controller decomposition and also incorporating hysteresis at switching points is utilized to smoothly transfer control authority between several MIMO controllers. The approach is applied to a high-fidelity truth-reference GT-Power engine model for a lean-burn natural gas-fueled engine to evaluate the closed-loop controller performance. The multi-tracking control problem targets engine speed, differential pressure across throttle as well as air-to-fuel ratio to achieve satisfactory engine performance and emissions without compressor surge. The engine response obtained using the robust MIMO controller is compared with that obtained using a state-of-the-art benchmark controller to evaluate the additional benefits of the MIMO controller.

This paper describes the design, algorithm development, and experimental validation of a novel LiDAR-based benchmark system to evaluate the in-situ performance of combine harvester grain unloading on-the-go automation systems that incorporate stereo camerabased perception. To find the appropriate sensor for the benchmark system, different LiDAR sensors were compared based on their field-of-view and performance in the presence of dust. A data processing framework was developed for the benchmark perception system to extract the reference grain fill map from the raw LiDAR point clouds via data preprocessing, cart boundary detection, cart tarp removal, and fill map generation. The LiDARbased benchmark system provides an accurate reference for the in-cart grain profile estimation with a height measurement error of less than 0.03 m. During unloading-on-the-go, this benchmark system runs simultaneously with the stereo-camera based system to provide the benchmark data via post-processing from in-field testing. The comparison provides quantitative analysis of the stereo-camera based system including the spatial distribution and the temporal progression of the measurement error in different unloading scenarios. Experimental results demonstrated that the proposed benchmark system provides consistent benchmark data with over 80% perception coverage of the grain cart in both clear and dusty environments. & COPY; 2023 IAgrE. Published by Elsevier Ltd. All rights reserved.

This paper describes a comprehensive framework for the development and validation of a model-based robust multivariate controller, which is used to regulate the gas exchange processes of an advanced turbocharged natural gas genset engine. The natural gas genset engine involved in this study features a multi-input and multi-output structure and is highly nonlinear, making it a challenge for which to design a controller. A control design-oriented model describing the dynamics of the studied engine is first developed and validated, using modeling techniques commonly employed for modeling of turbocharged engines. A novel robust model-based multivariate controller is then synthesized to incorporate the significant coupling between the engine actuators (throttle valve, bypass valve, and fuel valve) and the desired control objectives (engine speed, throttle differential pressure, and air-to-fuel ratio). To verify the effectiveness of the synthesized controller, it is compared to a benchmark production controller using an experimentally-validated simulation model. Simulation results indicate the merits of the synthesized robust multivariate controller in terms of better tracking performance (robust multivariate controller reduces steady-state error of engine speed, throttle differential pressure, and AFR at most by 100%, 58.9%, and 100% respectively, compared to benchmark production controller), faster transient response (robust multivariate controller reduces settling time of engine speed, throttle differential pressure, and AFR at most by 74.5%, 47.8%, and 71.2% respectively, compared to the benchmark production controller), and finer coordination among multiple interacted actuators when faced with multifacted control objectives.

Since the introduction of diesel urea SCR technology, aftertreatment thermal management has become critical for maintaining SCR catalyst light-off and thereby low cumulative cycle NOx emissions. A novel diesel engine aftertreatment thermal management strategy is proposed which utilizes a 2-stroke breathing variable value actuation strategy to increase the mass flow rate of exhaust gas. Experiments showed that when emissions are constrained to the same level as a state-of-the-art thermal management strategy, 2SB does not increase heat transfer to aftertreatment. However, if constraints are allowed to flex, temperatures comparable to a state-of-the-art thermal management calibration could be achieved with a 1.75× exhaust mass flow rate, potentially helping heat the SCR catalyst in a cold-start scenario.

Accurate modeling and control of the gas exchange process in a modern turbocharged spark-ignited engine is critical for the control and analysis of different control strategies. This paper develops a simple physics-based, five-state engine model for a large four-stroke spark-ignited turbocharged engine fueled by natural gas that is used in variable speed applications. The control-oriented model is amenable for control algorithm development and includes the impacts of modulation to any combination of four actuators: throttle valve, bypass valve, fuel rate, and wastegate valve. The control problem requires tracking engine speed to provide propulsive power, differential pressure across the throttle valve to prevent compressor surge, air-to-fuel ratio to restrict engine emissions. Two validation strategies, open-loop and closed-loop, are used to validate the accuracy of both nonlinear and linear versions of the control-oriented model. The control models are able to capture the engine dynamics within 5%–10% error at most of the engine operating points. Finally, the relative gain array (RGA) is applied to the linearized engine model to understand the degree of interactions between plant inputs and outputs as a function of frequency for various operating points. Results of the RGA analysis show that the preferred input-output pairing changes depending on the linear plant model as well as frequency. Therefore, a coordinated controller is ideal to tackle the control problem in question.

This paper describes the development and experimental validation of a novel grain unloading-on-the-go automation system (automatic offloading) for agricultural combine harvesters. Unloading-on-the-go is desirable during harvest, but it requires highly-skilled and exhausting labor because the combine operator must fulfill multiple tasks simultaneously. The automatic offloading system can unburden the combine operator by automatically monitoring the grain cart fill status, determining the appropriate auger location, and controlling the relative vehicle position and auger on/of. An automation architecture is proposed and experimentally demonstrated to automate the unloading-on-the-go process. To allow for diferent operator-selected unloading scenarios, the automatic offloading controller has three fill strategies and two movement control options, “open-loop” and “closed-loop”. The automatic offloading controller was implemented on a dSPACE MicroAutoBox II and integrated into a combine harvester. In addition, a stereo-camera-based perception system was connected to the automatic offloading controller via an Ethernet cable for grain fill profile measurement during unloading. In-feld testing demonstrated that the automatic offloading system can effectively automate the unloading-on-the-go of a combine harvester to fill a grain cart to the desired level under nominal harvesting conditions.

This paper describes the development, simulation, and experimental validation of a novel grain unloading-on the-go automation system (automatic offloading) for agricultural combine harvesters. Unloading-on-the-go is desirable during harvest, but it requires highly-skilled and exhausting labor because the combine operator must fulfill multiple tasks simultaneously. The automatic offloading system can unburden the combine operator by automatically monitoring the grain cart fill status, determining the appropriate auger location, and controlling the relative vehicle position and auger on/off. An automation architecture is proposed and experimentally demonstrated to automate the unloading-on-the-go process. To simulate the automatic offloading operation, a grain fill model and vehicle dynamics models were developed and validated with in-field testing. To allow for different operator-selected unloading scenarios, the automatic offloading controller has three fill strategies and two movement control options, "open-loop " and "closed-loop ". Simulation results demonstrated that both movement control options can achieve the fill target. The automatic offloading controller was implemented on a dSPACE MicroAutoBox II and integrated into a combine harvester. A PC-based user interface was developed for the combine operator to monitor unloading status and provide commands during the test. In addition, a stereo camera-based perception system was connected to the automatic offloading controller via an Ethernet cable for grain fill profile measurement during unloading. In-field testing demonstrated that the automatic offloading system can effectively automate the unloading-on-the-go of a combine harvester to fill a grain cart to the desired level under nominal harvesting conditions. The achievable fill level for a 1000-bushel grain cart without spillage ranges from -0.7 m to -0.2 m for the near-edge grain height relative to the cart edge.